Q: What kinds of well logs does Petro.ai use as inputs?
A: “The answer is that actually we can take in any logs,” Dr. Brendon Hall, VP of Geoscience explains, “It’s a popular data type that contains all kinds of information. The Petro.ai system is architected so we can add in new data types as they become available.
“You can take what we do with logs right now and divide it into two categories. One is to give Petro.ai insight into the geomechanical situation in the earth. They let us build the geomechanical model. The other one is to let us know more about the geological properties of the subsurface, or the petrophysical properties of what those logs measure.
“The geomechanical is very important for people working in unconventionals. Dr. Mark Zoback has developed a log based approach for determining stress profiles. There are a number of logs that are very useful for building that stress profile. Clay plus TOC, for example, tells you what fraction of ductile material is in the ground.
“Those clay plus TOC logs let you know how that material can deform over time. Dr. Mark Zoback developed something called the Viscoplastic Stress Relaxation theory, VSR. VSR states that the more a solid relaxes over time, the more SHmin, the minimum horizontal stress, or the stress that fractures need to overcome to propagate, can increase. Where you have areas with high clay and TOC you could have a stress barrier there and that’s important to know.
“Getting out of the geomechanics realm and entering geophysics, the oil saturation log is another one that’s very important. This is a tool that goes down and measures the amount of oil, the percentage of hydrocarbons that are at the various depths of the subsurface. The higher the oil saturation, the more oil and gas there is to exploit there.
“Sonic logs are also informational. They’re good for getting rock mechanical properties like Young’s modulus. It’s all giving you an idea of the subsurface which you can use to enhance your geologic model.
“One of the things we do with the logs is, we combine it with the frac fingerprint to derive the amount of drainage from a stimulation.”
“Right. Our main goal with the logs,” Kyle LaMotta, VP of Analytics adds, “is that they characterize stress as accurately as possible because that’s going to have a big effect on the drainage. If the stress is accurate, the drainage will be more accurate, and the overall model accuracy will improve. We definitely want to get that right, but then there are other things that might come up that our clients might have a question about, like the correct landing position.”
LaMotta continues, “And we can use proxies for these logs. One thing we use as a geologic feature proxy is lat and long. We know that the geology is changing spatially so the only reason to use a coordinate as an input to the model is to capture that change in the geology spatially. If they have a map of porosity and permeability or oil in place or GOR, we could use those as those would be better features that would explain more than lat and long. We’ll use those as place holders.
“The other thing we do for relative well comparisons is classify the landing interval. Based on their TOCs that they’ve picked for each interval, each well has an assigned landing interval which is also a proxy for the reservoir that it’s landing in. The model should be able to pick up on the relative performance of one interval versus another.
“The key takeaway is Petro.ai can incorporate lots of different logs. We can use them as features and we can interpolate them spatially and stratigraphically. We can take one log and apply it over a large area and include that in the model and apply that information to different wells. We use key geomechanical logs in the model and we can also incorporate geophysical and geological data types as logs too.”
